A confidentiality event is an incident where sensitive information is exposed, disclosed, or made accessible to someone who should not see it, without necessarily altering or destroying the data itself. The core problem is loss of secrecy rather than loss of availability or integrity. This matters because attackers often prefer quiet disclosure over disruptive damage: stolen mail, documents, credentials, or private AI context can be monetized or reused without triggering obvious alarms.
In practice, confidentiality events can happen through phishing, misconfigured access controls, prompt injection, unsafe sharing links, or abuse of trusted services that already have permission to read data. Defenders look for unusual access patterns, overbroad permissions, and unwanted data retrieval by assistants or applications. Controls such as least privilege, service separation, content filtering, and user awareness reduce the chance that a legitimate session becomes a data leak.